From Dataset
to Finished Solution.
I work with large datasets, develop deep learning models and deploy them with Docker and CI/CD – complemented by modern agentic AI approaches such as RAG and prompt engineering.
What I bring
My strengths range from data processing and neural network training through classical optimisation to deployment and AI-assisted development.
Large Datasets & Visualisation
Processing and visualising very large scientific datasets with efficient I/O and clear presentation.
Deep Learning
Developing and training neural networks and classical ML models for complex prediction and classification tasks.
Optimisation Algorithms
Applying classical numerical and combinatorial optimisation methods to solve complex engineering and research problems.
Deployment & CI
Containerising applications and setting up automated continuous-integration pipelines for reproducible deployments.
Agentic AI
Building autonomous AI agents and RAG systems; prompt engineering and AI-assisted development with Claude Code.
Interdisciplinary Collaboration
Bridging technical and non-technical teams: understanding requirements, communicating solutions clearly, and delivering together.
1. Shaping data→2. Developing systems→3. Shipping products
Selected Projects
From a production RAG system to research projects in deep learning and optimisation – a glimpse into my work.
Paragraphenreiter
Demo →
RAG system for legal questions: user input is semantically analysed and automatically linked to the relevant legal paragraphs.
METRIX Beamline Offset Detection Code →
Offset finding for the METRIXS beamline at BESSY II using a deep learning surrogate model combined with a genetic algorithm.
SRF Electron Gun RL Optimization Code →
Deep reinforcement learning system for automated optimisation of the SRF electron gun parameters at Sealab (Helmholtz-Zentrum Berlin).
Time-of-Flight Signal Reconstruction Code →
Deep learning approach for denoising and reconstructing time-of-flight detector data with missing or corrupted readings.
X-ray Pole Figure Reconstruction Code →
Reconstruction of incomplete X-ray diffraction pole figures of oligocrystalline materials using neural networks.
Ready for the
next project?
Got a Project in Mind?
I look forward to interesting challenges around agentic AI and RAG.
Send an email